9 min read
You've seen the story. Computer science graduates can't find work while philosophy majors are being hired to think for a living. It's a satisfying reversal and it gets shared constantly.
It's also about a third true. The tech slump versus philosophy boom framing contains a real signal wrapped in a lot of narrative, and if you make a career decision on the narrative you'll regret it. This piece separates what's actually observable from what's a good headline, and says what the underlying pattern means for you either way.
Whatever you conclude, the practical bottleneck is the same. Run your file through the free ATS resume checker , because in a competitive market screening gets stricter, and a file that parses badly fails regardless of which side of this debate you're on.
Quick Wins
- Count actual postings for the role you're considering. Trends don't hire, employers do.
- Write one bullet showing judgement, not just output: what you decided and why.
- Name the domain you understand deeply. That's the half everyone forgets.
The claim, stated fairly
The argument goes roughly like this. Software that writes software has compressed demand for junior technical labour, while the same technology has created demand for people who can judge, critique, and reason about outputs. Therefore: fewer coders, more philosophers.
Two things are worth noticing before you accept or dismiss it.
First, it's a story with an appealing shape. Educated people who studied humanities have spent two decades being told their degrees were worthless, and a reversal narrative is emotionally satisfying regardless of its accuracy. Satisfying stories travel further than accurate ones.
Second, the two sides of it are not comparable in size. Even if demand for reasoning-heavy roles is genuinely rising, it's rising from a small base. A doubling of a small number is still a small number, and headlines rarely say so.
What's true about the tech side
The tech slump half has more substance to it, with one important correction: it isn't uniform, and treating it as uniform is what makes the advice useless.
The pattern visible in postings has been a split rather than a collapse. Roles tied to infrastructure, security, data platforms, and deploying AI systems into real organisations have behaved very differently from general entry-level application development. Same industry label, opposite candidate experience.
The entry-level squeeze is the genuinely hard part, and it isn't only about automation. Junior roles are also the first thing cut when budgets tighten, because the productivity return is slowest. That happens in every downturn and it looks identical to technological displacement from the inside.
The practical takeaway: don't ask whether tech is hiring. Ask whether your specific job family is hiring in your specific metro, and count the postings to find out.
What's true about the humanities side
A real category of work has appeared that rewards exactly what a good humanities education teaches: constructing an argument, finding the hole in one, writing with precision, and being comfortable with ambiguity.
Where those roles actually exist:
- AI evaluation and red-teaming. Judging whether an output is good, wrong, or subtly misleading.
- Policy and governance. Deciding what a system should and shouldn't do, and writing that down defensibly.
- Trust and safety. Applying rules to messy real cases where the rule doesn't quite fit.
- Technical writing and documentation. Explaining a complicated thing accurately to someone who needs to act on it.
- User research. Working out what people actually need rather than what they said.
Now the caveats, which the enthusiastic version of this story leaves out. Most of these postings ask for domain knowledge alongside the reasoning ability. Many are contract or project-based rather than permanent. And the number of them, nationally, is modest.
"Philosophy graduates are in demand" is a stretch. "People who can reason carefully and also understand a domain are in demand" is closer, and it's a more useful sentence because it tells you what to build.
The pattern underneath both
Strip the narrative away and something more durable is visible. The split isn't technical versus humanities. It's producing versus deciding.
| Under pressure | Holding up |
|---|---|
| Producing a predictable output to a clear specification | Deciding what the specification should be |
| Summarising or reformatting information | Judging whether the summary is right and what to do about it |
| Routine work with no context required | Work needing deep context about one messy domain |
| Tasks where being wrong is cheap | Tasks where being wrong is expensive and someone must own it |
Notice that this cuts across both fields. A software engineer designing a system nobody has specified is on the right-hand side. A philosophy graduate writing generic content is on the left. The degree isn't doing the work; the nature of the task is.
That's also a more actionable frame, because you can move along it without changing career. Take on the ambiguous problem instead of the well-specified one, and own the decision rather than just the output.
What to do, from either side
If you're technical
Don't abandon the field on the strength of a narrative. Move toward the parts of it that are demonstrably hiring, and toward work where you decide rather than implement. Pick a domain and learn it properly, because an engineer who genuinely understands claims processing or grid operations is far harder to replace than one who is fluent in a framework.
If you're from a humanities background
Your reasoning is real and it is not sufficient on its own. The roles that reward it nearly always want a domain attached: healthcare, finance, law, a specific product area. Add the domain, and make the reasoning visible on paper rather than implied by your degree.
Showing judgement on a resume
This is the part almost everyone gets wrong. Judgement doesn't appear in a responsibilities list. It appears when you name a decision.
| Output framing | Judgement framing |
|---|---|
| Wrote content policy documentation. | Decided where the moderation line sat on three contested categories, and wrote the policy the review team now applies. |
| Reviewed model outputs for quality. | Built the evaluation rubric that separated confidently wrong answers from uncertain ones, cutting reviewer disagreement by half. |
How to check the claim yourself
You don't have to take anyone's word for this, including mine. The test takes about fifteen minutes and it's worth more than every trend piece on the subject.
- Pick the exact job title you're considering, not the field. "AI policy analyst," not "philosophy jobs."
- Search it filtered to the last week, in your metro and then remote, and count the results.
- Open five of them and read what they actually require. Note how many ask for domain experience alongside the reasoning ability.
- Do the same for your current title as a baseline. The comparison is the useful output, not either number alone.
- Repeat monthly. One count is a snapshot; three counts is a direction.
People are often surprised by how small the counts are on both sides. That's the real answer to this debate, and it's available to anyone willing to spend a quarter of an hour instead of reading another article about it.
Edge cases
You're choosing a degree now
Choose on a longer horizon than the current cycle, because a degree takes years and the hiring conditions you graduate into will not be today's. Combining a domain with strong writing and reasoning has aged better than either alone.
You're a laid-off engineer
Retraining into humanities work is a much longer road than moving sideways within technical work toward a healthier job family or a different industry. Check local postings by job family before making a large decision.
You want an AI evaluation role
Many are contract or hourly, and the market for them is genuinely competitive. Treat one as a foot in the door rather than a destination, and use it to build the domain knowledge that makes the permanent version possible.
You're mid-career with neither profile
You may be better placed than either group. Years of domain knowledge is exactly the half that both sides of this debate are missing, and it's the part that can't be acquired quickly.
Your work is genuinely being automated
Move toward the ambiguous end of your own field rather than to a different field. The tasks nobody has specified yet are the ones that stay, and you can usually reach them from where you already stand.
Mistakes this narrative causes
- Making a multi-year decision on a headline. By the time a trend is a trend piece, it's already partly priced in.
- Assuming a degree is the qualification. Employers hire the demonstrated skill. The degree is evidence, not the thing itself.
- Treating tech as one market. The variation inside it is larger than the difference between sectors.
- Leading with reasoning and no domain. The postings almost always want both, and the domain is the harder half to fake.
- Writing a resume full of outputs. If nothing on the page names a decision you made, nobody can assess your judgement.
- Waiting to see how it plays out. Whatever the answer, applying with a current, well-targeted file beats watching.
Make the case on paper
Whichever side you're on, the same problem decides your search: whether a screening system and a stranger can tell what you're actually good at in six seconds.
Upload your resume to the free ATS resume checker and read the extraction. Judgement-heavy work is especially easy to under-sell, because the interesting part was a decision and the resume only records the deliverable. Rewrite the top three bullets so the decision is visible.
Then aim carefully. The job match score shows whether a posting genuinely fits before you spend an hour on it, and the cover letter generator is where a cross-field application explains itself, which a resume alone cannot do.
The short version
- The tech slump versus philosophy boom framing has a real signal in it and a lot of narrative around it.
- The durable split is producing versus deciding, and it cuts across both fields rather than between them.
- Reasoning plus a domain beats either alone. Add the half you're missing.
Do this today: rewrite one resume bullet so it names a decision you made and why, rather than the thing you produced.
Then check it lands. Check your resume for free and see whether your judgement is visible or buried.
Read more
- Which industries are actually hiring — how to check rather than guess.
- 7 habits to future-proof your career — what holds regardless of the trend.
- How to beat AI resume screeners — the gate that decides both sides of this.
Frequently asked questions
There is a real and growing set of roles that reward reasoning, judgement, and careful writing, particularly around AI evaluation, policy, and trust and safety. Calling it a boom overstates it, because the absolute number of those jobs is small compared with the fields people are being told to leave.
Tech hiring has been uneven rather than uniformly weak. General entry-level software roles have been much harder to get than infrastructure, security, data platform, and AI deployment work. Treating tech as a single market produces advice that fits nobody.
No degree is a job guarantee, and choosing one purely on a hiring trend is risky given how long a degree takes. What transfers is the underlying skill: constructing an argument, spotting a flawed one, and writing precisely. Those can be demonstrated without the credential.
AI evaluation and red-teaming, policy and governance, trust and safety, technical writing, user research, and communications roles inside technical organisations. Most of them ask for domain understanding alongside the reasoning ability, not instead of it.
Work that combines judgement with a domain. Roles where the task is deciding what should happen, in a context you understand deeply, have held up better than roles where the task is producing a predictable output to a specification.